After much speculation and an official unveiling last week, it's finally time to see if Threadripper can bring competition to the high-end desktop segment while delivering the value and efficiency we've come to expect from other Ryzen processors.
Before we jump into the benchmarks, here's how we've set up the comparison. For this review, we'll be testing the new Threadripper 1950X and 1920X along with the Ryzen 7 1700 and 1800X, as well as Intel's competing Core i7-7800X, 7820X and Core i9-7900X.
Compared to the Ryzen 7 1800X, which until now was leading AMD's pack, the Threadripper 1920X comes at a 75% price premium and offers 50% more cores. You also get quad-channel memory support and significantly more PCI Express lanes.
While we're curious to see how Threadripper fares against AMD's more affordable lineup, I'm more interested in learning how they compare against Intel's Skylake-X parts, so you can expect an emphasis on those results.
The Skylake-X CPUs were installed on an Asrock Fatal1ty X299 Professional Gaming i9 motherboard, while the Threadripper CPUs were tested on the Asrock X399 Taichi motherboard. Both platforms were fitted with G.Skill's TridentZ DDR4-3200 CL14 64GB quad-channel memory kit. Big thanks go to G.Skill for providing all that memory.
The X299 platform was installed on the Praxis Wetbench and was cooled using Thermaltake's Pacific RL360 custom liquid cooled kit. Meanwhile Threadripper was tested on a sheet of foam on my desk and was cooled by the Thermaltake Floe Riing RGB 360 AIO liquid cooler.
Ryzen Threadripper System Specs
Skylake-X System Specs
Ryzen 7 System Specs
First up, memory bandwidth performance. As you probably know, Ryzen 7 CPUs feature a dual-channel memory controller whereas Threadripper offers quad-channel memory access, much like Skylake-X. As a result, memory bandwidth has been increased by 50% over Ryzen 7 and now Threadripper is on par with the Core i9-7900X.
As promised by AMD, the 1950X is good for an incredible multi-threaded score of 3028pts in Cinebench. That's a 39% increase over the i9-7900X, both of which cost $1,000.
The Core i9 CPU is 10% faster for single-threaded tasks, but that's not exactly the point of these CPUs. The 1920X was also 13% faster than the 7900X and 40% faster than the i7-7820X, so a clear win here for AMD in terms of price vs. performance.
Before we move onto the more serious benchmarks, I wanted to see how Threadripper stood up in PCMark 10, which looks at office type workloads and typically isn't core-heavy. To my surprise, Threadripper scored very well in this test and did considerably better than the Core i9-7900X.
Excel can utilize many threads, especially when running the extreme Monte Carlo simulation. Here the 1920X just edged out the 7900X, taking an impressive 1.86 seconds to complete the workload. That said, it was 12% slower than the 1950X which took just 1.66 seconds making it the fastest desktop CPU we've ever tested here.
For those of you unaware, VeraCrypt is an open-source utility used for on-the-fly encryption and features optimized implementations of cryptographic hash functions and ciphers which boost performance on modern CPUs. It also supports parallelized encryption for multi-core systems as well hardware-accelerated AES to further improve performance.
In short, it takes full advantage of the many cores Threadripper offers and we see that here. Compared to the Core i9-7900X, the 1950X was 48% faster in the 50MB test and 35% faster in the 1GB test. The 1920X also easily beat the 7900X delivering up to 25% more performance.
Next up we have unencrypted compression and decompression performance using 7-Zip. Whereas Hyper-Threading sees a similar level of efficiency when compressing and decompressing, SMT is significantly more efficiency for decompression work. SMT boosts the performance of Threadripper by around 15% for compression work, whereas I saw a massive 60% increase for decompression. Disabling SMT saw the 1950X deliver around 50,000 MIPS for both tests.
However, with SMT enabled the 1950X just beat the Core i9-7900X for the compression test but crashed it by a 53% margin for the decompression test. Meanwhile the 1920X matched the compression performance of the 7820X but was 48% faster when comparing the decompression results. Overall very strong results here for AMD's new Threadripper CPUs.
Handbrake is a popular application for encoding video and we've used it to convert a 4K H.264 video to 1080p using H.265 and recorded the average frame rate. The 1950X was 6% faster than the 7900X in this test, not a huge gain but given both CPUs cost the same amount it's a good result for the Threadripper CPU. That said the 1920X does cost 33% more than the 7820X and yet it was just 5% faster here, so a less impressive result for AMD in that comparison.
The Premiere Pro CC results interest me the most as I spend a lot of time rendering these 4K videos on my Core i7-6950X editing machine. Back when I first reviewed the Core i9-7900X I said that I planned to probably replace my 6950X with Threadripper.
However, I also noted that the current version of Premiere Pro CC isn't that good at taking advantage of high core count CPUs and as a result the 7900X was just 4% faster than the 6950X, so a bit of a disappointing result there. This benchmark is CUDA-accelerated by a GTX 1080 Ti so keep that in mind, but there aren't a huge amount of effects in my videos for the GPU to accelerate.
Given that the higher-clocked 7900X was just 4% faster than my 6950X and both are 10-core parts, I wasn't hopeful for Threadripper to save me much time. Needless to say, I was pleasantly surprised when the 1950X took just 128 seconds to render the 1 minute 30 seconds sample video. That's a 12% improvement over the 7900X. Meanwhile the 1920X was also just 7% slower than the 7900X.
All of this is to say that it looks indeed like my next build will employ a Threadripper processor -- the 1950X specifically.
Moving on to Blender's Ryzen Graphic 27 test, the 1950X offered a massive 29% performance improvement over the 7900X, and in fact even the 1920X bested the Core i9 CPU, albeit by a slim margin, but a solid result for AMD nonetheless. Let's move on to check out a more extreme Blender workload.
The Gooseberry workload takes some serious firepower to complete in around half an hour and we see this as the 1950X turns in a time of almost 32 minutes but that's a decent improvement on the 36 minutes it takes the Core i9-7900X, a 13% improvement in fact. This time the Core i9 CPU does just beat the 1920X though the margin was extremely slim.
Corona comes as a standalone benchmark. It renders a fixed scene six times and we take the time it takes to complete the task. This application loves threads to the more the merrier and the quicker you'll be done and can move on to something else. Here we see yet again that the 1920X is able to best the Core i9-7900X, even it is by a slim margin. Meanwhile the 1950X races ahead smashing the 7900X by a convincing 31% margin.
POV-Ray is another ray-tracer and it's been around for many years. When testing with the official benchmark, the 1920X bests the i9-7900X by a rather convincing 10% margin -- AMD's best result yet. Meanwhile, the 1950X was a whopping 37% faster than the Core i9 CPU and for those wondering almost 90% faster than the 1800X. This concludes the productivity benchmarks, time for a few quick games.
Before we get to the gaming benchmarks, be aware that there are two different memory access modes for Threadripper and they impact gaming performance quite a bit. By default, the CPU is configured to use the distributed mode which uses a UMA or Uniform Memory Access configuration. This method works best for the productivity workloads we just saw and since it's the default configuration I decided to test the games using this mode. That said, I will compare UMA vs. NUMA performance as well.
First up we have Battlefield 1 and although performance doesn't look great, it doesn't look bad either with the Threadripper CPUs delivering similar performance to the R7 1700. That does however make them quite a bit slower than the Skylake-X CPUs, particularly the 7820X and 7900X. Please note that for our gaming tests I've lowered the quality settings slightly to try and reduce the GPU bottleneck.
Moving to Total War: Warhammer we see some pretty ordinary performance, particularly in relation to Ryzen 7. It's possible the game will be updated to better support Threadripper, that said though we are pushing over 100fps at all times so while comparatively slow when compared to the 1800X and 7900X, Threadripper's not exactly slow.
The 1950X performed well when testing with Ashes of the Singularity: Escalation, basically matching the Core i9-7900X, while 1920X wasn't that far off, beating the 7820X. Both Threadripper CPUs were much faster than the Ryzen 7 models.
Ryzen 7 has proven to be very fast in Civilization VI and while not as fast, Threadripper also performs well in this title, beating out the Intel CPUs.
The last game we have time for is F1 2016 and here Threadripper slips behind Ryzen 7 and well behind Skylake-X. With the quality preset reduced to high we are seeing some pretty massive frame rates from the Intel processors.
UMA vs. NUMA
As mentioned on the last page, you switch between memory access modes on Threadripper processors. 'Distributes' mode uses 'Uniform Memory Access' is enabled by default so this is what we've been testing with so far. However, you can switch to 'local' mode for 'Non-Uniform Memory Access' and this allows each of the Zeppelin dies to prioritize which cores access certain parts of the system memory. This basically prioritizes the nearest cores to improve overall latency for gaming applications that tend to place a premium on fast memory access. So let's see how this impacts gaming performance as well as a few productivity workloads...
Switching to NUMA has improved the Battlefield 1 performance and now Threadripper is delivering similar frame rates to the Ryzen 7 1800X. In fact, the 1950X creeps ahead ever so slightly and now is that much slower than the Core i7-7820X and Core i9-7900X.
F1 2016 also sees decent gains when switching to NUMA though this time isn't able to match the Ryzen 7 1800X and this means it's also well down on the Intel CPUs.
Performance in Civilization VI was already great and no extra performance was had when switching to NUMA.
Looking at application performance, we see a slight performance decline with POVRay, but nothing serious.
Blender also saw a slight decline in performance with the 1950X being about 4% slower using NUMA.
Last up we have the Corona test, in which both NUMA and UMA offered the same performance. There's not much to see here, so let's move on to some consumption figures.
Power & Temps
Power consumption is one area where AMD continues to impress. Come to think of it, Intel has also been quite surprising in this department lately, but for the opposite reasons. Whereas the 10-core 7900X pushed total system consumption to an incredible 281 watts in the Corona benchmark, the 16-core 1950X reached just 257 watts, a 9% decrease in total consumption.
That's a truly impressive result for AMD given that the 1950X was faster in every single one of our productivity workloads, sometimes by quite a significant margin, and yet it consumes less power.
The 1920X was extremely fuel-efficient, consuming 26% less power than the 1950X and just 18% more than the 1800X. It also somehow use 15% less power than Intel's 6-core 7800X.
Let's see how things heat up once overclocked...
Using the Thermaltake Floe Riing RGB 360 TT Premium Edition AIO liquid cooler (that's a mouthful), we found some solid temperatures with Threadripper. Using the factory installed TIM, the Floe was fitted to the 1950X with the CPU overclocked to 4.0GHz for all 16-cores. The system ran at a stress temperature of 88 degrees which I thought was good given what we've seen recently from the Core i9-7900X.
Threadripper 1950X @ 4 GHz (All Cores) Stock TIM
Threadripper 1950X @ 4 GHz (All Cores) Reapplied-TIM
Threadripper 1950X (Stock) Reapplied-TIM
However, after removing the cooler and cleaning the TIM, we covered the entire copper base in paste and tried again. To my surprise, temperatures were reduced quite drastically, now peaking at 83.5 degrees but often sitting at just 81 degrees.
With the custom paste still applied, I reverted back to the stock out of the box clock speeds and now the 1950X was hitting just 71 degrees.
For those wondering, temperatures were much the same for the 1920X under the same conditions.
As expected the Threadripper CPUs were able to hit 4 GHz but unfortunately like Ryzen 7 weren’t able to go beyond that point. Extreme voltages might see you hit 4.1 - 4.2 GHz like what we’ve seen with some Ryzen CPUs but for 24/7 usage 4 GHz looks to be the limit.
Anyway at 4 GHz the 1950X spat out an incredible score of 3408 pts in Cinebench though due to the way XFR works this overclock actually slightly reduced the single thread performance. You could probably fine tune the overclock to avoid this but with time against us we didn’t play around with the overclocking too much.
Overclocked the Blender render time for the 1950X was cut down by an impressive 11% while the 1920X time was reduced by 6%. This means even overclocked the Core i9-7900X wasn’t much faster than the 1920X and wasn’t able to match the stock 1950X in this application.
Interestingly this time when testing with Corona the 1920X saw a massive 17% performance improvement when overclocked while the 1950X was just 7% faster. This meant both overclocked Threadripper parts were much faster than the overclocked 7900X.
Back to the kind of margins seen in Blender we find that while the 1950X was 9% faster overclocked the 1920X was just 5% faster. This meant when testing with Premiere Pro CC the 1920X did trail the Core i9-7900X while the 1950X was quite a lot faster.
Overclocking performance in Warhammer was tested using Uniform Memory Access and here the overclocks did little to boost performance.
Finally we have the overclocking power consumption figures and here Threadripper looks quite good, especially in relation to the Core i7-7900X. Whereas the Intel CPU pushed total system consumption to 388 watts, the 1950X hit 358 watts, an 8% reduction for what was often between 20 - 30% more performance.
Price vs. Performance
To compare price vs. performance I've taken the current retail cost for each processor from Amazon and you can find those links at the bottom of this page.
When looking at Blender, the Threadripper 1950X delivers the best performance as it's situated the furthest right on our scatter plot and as you can see it also sits lower than the Core i9-7900X since it's also cheaper.
To give you a better idea of what's going on, here is a linear guide from the Ryzen 7 1700 to the Threadripper 1950X. CPUs positioned left of this line represent poorer value while anything to the right is better in terms of price vs. performance.
This gives us a good idea how things look between AMD's cheapest 8-core CPU and its new flagship parts. As you can see, the Intel Core i7-7800X and 7820X are very competitive while the 1920X offers the best overall value. The only CPU to deviate a notable distance from the line is the Core i9-7900X and since it's to the left, we find it to be a poor value.
Next up we have Corona and this benchmark looks to be largely dominated by AMD and all four of their CPUs deliver a similar price vs. performance ratio. Intel's best value offering here is the Core i7-7820X while the 7800X and 7900X represent a rather poor value.
The POVRay price vs. performance scatter plot looks remarkably similar to what we saw with Corona. Again we see that all four AMD processors provide a similar value, while only the Core i7-7820X looks half decent here for Intel while its 7800X and 7900X are a bad value.
Finally, I wanted to check out price vs. performance in Premiere Pro CC. The AMD processors look to be ahead in terms of value, while the Core i7-7820X seems like the best buy out of Intel's Skylake-X CPUs. The overpriced 7900X again represents a poor value.
Core i9 Killer: Who Is It For?
That was a lot of data but it all indicated one thing: total domination by AMD's new Threadripper processors. Their only weakness was gaming but there's really no point in buying Threadripper for that. If your goal is great gaming performance and you have $1,000 to spend on a processor, get the $340 Core i7-7700K and pocket the change for the best GPU you can afford. The 7700K is significantly cheaper and much better suited for gaming.
The reason I've recommended the Ryzen 5 1600 over the i7-7700K in the past is because it's a much cheaper part ($215) on a more affordable platform, and it will deliver the same experience for 90% of gamers out there. So again, while the 7700K is the best option if you want the ultimate gaming experience with maximum frame rates, Ryzen makes more sense in terms of value.
So Threadripper may not be ideal for gaming but it's still capable and it's likely much better than Intel's Xeon Gold 6130 for example, a $1,900 16-core/32-thread server chip that operates at a base frequency of only 2.1GHz.
But enough about gaming, as that clearly isn't the point of these 12-core and 16-core CPUs. Productivity is what matters here and for serious workloads Ryzen has already proven to be a beast. Threadripper takes things to a new level, though.
We now have a $1,000 processor from AMD and compared to the competition it's actually worth the asking price. Thousand dollar desktop processors have been around for some time now, but it's been a while since we've seen such an expensive solution from AMD. Threadripper certainly offers the best bang for your buck in this category.
Looking back at Broadwell-E, Haswell-E, Ivy Bridge-E and so on, Intel has been hands down the best option for power users for the simple fact that they were the only choice. Intel has won by default over the past five-plus years.
Back when Intel had no competition, the pricing of early Extreme Edition CPUs bothered me, but how can you complain when there's no alternative? By the time Haswell-E came around, we'd kind of accepted the $1,000 asking price for a flagship Intel part, which at the time was an 8-core/16-threaded chip. Then Broadwell-E came around and Intel couldn't imagine a world where they would return to facing stiff competition, so they went wild with pricing and we ended up with the outrageous $1,700 6950X.
Now, Skylake-X and the X299 platform are not bad. Yes, there are a few head-scratching aspects, but overall they're good products. The problem is with their price, or at least how that shakes out when you compare it to performance. Skylake-X has also improved value-wise, but as we said previously when comparing the 7820X and R7 1700, we've kind of just gone from stupid pricing to a little less stupid. Meanwhile, AMD has been more aggressive setting the Ryzen 7 beautifully, and Threadripper continues that strategy.
Because of pricing, core count becomes somewhat irrelevant: the Core i9-7900X and Threadripper 1950X both have a lot of cores and they both cost a grand. The key difference being that the 1950X often delivers 20-30% more performance while consuming almost 10% less power under full load.
In a nutshell, for the same price you get more CPU performance, lower power consumption and improved operating temperatures with Threadripper, so why would you buy the Core i9-7900X over the 1950X?
Intel recently announced the specs for its upcoming 12, 14, 16 and 18-core Skylake-X parts. At a guess, I would say consumers will need to spend at least $1,400 on the i9-7940X to match AMD's 1950X and I seriously doubt the 40% price premium will be worth it.
The X299 platform's lack of ECC memory support is another issue. Whereas Threadripper supports ECC, the Skylake-X chips don't and that means anyone serious about their workstation won't even consider Intel's high-end desktop platform. Unless Intel is willing to budge on pricing I can't see why anyone would invest in X299.
At the beginning of 2017, who would have thought we'd see AMD dethrone Intel at the very top of the high-end desktop CPU segment. It certainly wasn't us, but we're glad we can look forward to some competition. It's an exciting time to be a PC enthusiast.
Pros: The Threadripper 1920X offers more performance, uses less power and runs cooler than the Core i9-7900X. It also supports ECC memory (Intel's X299 platform doesn't). Compared to the Ryzen 7 1800X, the 1920X touts quad-channel memory support and more PCIe lanes.
Cons: The $1,000 Threadripper 1950X is competitively priced but nonetheless comes at a hefty premium. Threadripper falls a tad behind when it comes to gaming.